How HIV-1 entry mechanism and broadly neutralizing antibodies guide structure-based vaccine design

Abstract

Purpose of review

An HIV-1 vaccine that elicits broadly neutralizing antibodies (bNAbs) remains to be developed. Here, we review how knowledge of bNAbs and HIV-1 entry mechanism is guiding the structure-based design of vaccine immunogens and immunization regimens.

Recent findings

Isolation of bNAbs from HIV-1-infected donors has led to an unprecedented understanding of the sites of vulnerability that these antibodies target on the HIV-1 envelope (Env) as well as of the immunological pathways, which antibody lineages follow to develop broad and potent neutralization. Sites of vulnerability, however, reside in the context of diverse envelope states required for HIV-1 entry, including a prefusion-closed state, a single-CD4-bound intermediate, a three-CD4-bound intermediate, a pre-hairpin intermediate, and postfusion states, and it is not always clear which structural state optimally presents a particular site of vulnerability for vaccine elicitation. Furthermore, detailed knowledge of immunological pathways has led to debate among vaccine developers as to how much of the natural antibody-developmental pathway immunogens should mimic, ranging from only the recognized epitope to the entire antibody-virus co-evolution process.

Summary

A plethora of information on bNAbs is guiding HIV-1-vaccine development. We highlight consideration of the appropriate structural context from the HIV-1-entry mechanism and knock-in mice results showing extraordinary progress with replicating template B-cell ontogenies.

Introduction

The development of an effective vaccine remains a key challenge of HIV-1 research. Multiple groups have undertaken knowledge-based approaches with the aim of developing an effective B cell-based vaccine. These approaches seek information on two critical areas: (i) broadly neutralizing antibodies (bNAbs), which develop after 5+ years in a substantial proportion of patients infected by HIV-1 and are capable of neutralizing diverse strains of HIV-1 [1–5, 6••,7••], and (ii) the structure and conformations of the HIV-1 envelope (Env), a trimeric heterodimer comprising three gp120-exterior subunits and three gp41-transmembrane subunits, which is the sole target of virus-directed bNAbs (reviewed in [8•,9]).

Ground-breaking developments – involving diverse technologies including single molecule fluorescence resonance energy transfer (smFRET) [10], cryo-electron microscopy (cryo-EM) [11••,12••], X-ray crystallography [13••,14••] and nuclear magnetic resonance (NMR) [15,16] – are revealing the structures and conformations of the HIV-1 Env, a type 1 fusion machine that uses conformational change to drive fusion of viral and cellular membranes. These studies provide the context in which to situate bNAb sites of vulnerability. Meanwhile, insights from antibody-virus co-evolution [17,18•,19••,20••,21–24] involving next-generation sequencing (NGS) analysis of B cell transcripts and of evolving Env are now making their way into immunization efforts with germline targeting and knock-in mice [20••,25,26•,27••,28••,29••].

Here we review how insights from bNAbs and Env-entry mechanism are now being incorporated into HIV-1-vaccine immunogens and immunization regimens.

HIV-1 bNAbs

Early generation bNAbs, including b12, 2G12, 2F5, and 4E10 [30–33], exhibited limited breadth and potency yet they revealed a number of striking (and now known to be common) features of HIV-1 bNAbs. These include extensive somatic hypermutation [34,35] or extended heavy-chain third complementary determining regions (CDR H3s) [36], used to overcome barriers imposed by HIV-1 Env. Advances in B-cell technology with single memory B-cell sorting using epitope-specific probes [37,38] or direct neutralization screening [17,39,40] have led to identification and characterization of new bNAbs, which exhibit increased breadth and potency and target five conserved regions of vulnerability (Table 1).

Table 1.

Broadly neutralizing antibodies targeting HIV-1.

HIV-1 bNAb class	Representative Antibodies	Recognition properties	Epitope features	Isolation year	References
Apex (V1V2 glycan)
PG9 class	PG9, PG16	Long, protruding CDR H3 with sulfated tyrosine	N156/N160 glycans, V1V2 strand C	2009	[39,41,42]
PG9 class	CH01-CH04	Long, protruding CDR H3	N156/N160 glycans, V1V2 strand C	2011	[41,43,44••]
PGT145 class	PGT141-PGT145/ PDGM1400-1412	Long, protruding CDR H3 with sulfated tyrosine, quaternary specific	N160 glycan, hole at trimer apex	2011	[40,41,45,46••]
PG9 class	CAP256-VRC26.01-33	Long, protruding CDR H3 with sulfated tyrosine	N156/N160 glycans, V1V2 strand C	2014	[17,47]
N90-VRC38 class	N90-VRC38.01-11	Non-protruding, typical CDR H3, side chain interactions	N156/N160 glycans, V1V2 strands A, B and C	2016	[48]
N332 glycan patch
2G12 class	2G12	Domain exchanged structure	Glycan only	1994	[49]
PGT121 class	PGT121-123, PGT124/10-1074	25-residue CDR H3 with nonpolar tip	GDIR motif, V1V2 and/or V3 glycans	2011	[40,50–52]
PGT128 class	PGT125-131	Extended CDR H2 and CDR H3	GDIR motif, V3 glycans	2011	[40,53]
PGT135 class	PGT135-137	Extended CDR H1 and CDR H3	V3 and V4 glycans	2011	[40,54]
PCDN class	PCDN-27A, -27B, -33A, -38A, -38B	Extended CDR H3, low SHM and no indels	V3 glycans	2016	[18•]
PGDM class	PGDM11-14, PGDM21	Allosteric inhibition of CD4 binding	GDIR motif, N156 and V3 glycans	2016	[55•]
DH270 class	DH270.1-6	CDR H3 dependent	GDIR motif, V3 glycans	2017	[56•]
CD4-binding site
b12 class	b12	CDR H3 loop-dominated	CD4bs	1991	[30,57]
HJ16, CH103, VRC13, VRC16 classes	HJ16, CH103, VRC13, VRC16	CDR H3 loop-dominated	N276 requirement for HJ16	2010	[58,59]
VRC01 class	VRC01, N6, VRC23, VRC-PG04, 3BNC117	VH1-2 derived, 5-residue CDRL3, very high SHM	CD4bs	2010	[38,59–61,62••]
8ANC131 class	8ANC131, 8ANC134, 1B2530	VH1-46 gene derived, normal length CDR L3, very high SHM	CD4bs	2011	[59]
IOMA class	IOMA	VH1-2 derived, normal length CDR L3, low SHM	CD4bs	2016	[14••]
gp120-gp41 interface
PGT151 class	PGT151-158	Trimer and cleavage specific, asymmetric 2:1 (Ab:trimer) binding, YYYY motif in CDR H3	Fusion peptide and complex glycans at N611 and N637 on gp41	2014	[11••,63,64]
35O22 class	35O22	Binds cleaved and uncleaved Env	Glycan at N88 requirement	2014	[65,66]
8ANC195 class	8ANC195	Recognizes closed and open forms of Env trimer and gp120 monomer	Glycans at N234, N276, N637	2014	[12••,67,68]
3BC315/3BC176 class	3BC315/3BC176	Destabilizes trimer	gp41-gp41 interface	2015	[69]
N123-VRC34 class	VRC34.01-07	Trimer and cleavage specific	N-term of gp41 fusion peptide and glycan at N88	2016	[70••]
ACS202 class	ACS201-205	Trimer and cleavage specific, YYYY motif in CDR H3	gp41 fusion peptide and glycan at N88	2016	[71•]
CAP248-2B class	CAP248-2B	Cleavage dependent	Includes gp120 C-terminus	2017	[72•]
MPER
2F5 class	2F5	Extended CDR H3 with hydrophobic tip	N-terminal ELDKWAS core sequence	1994	[32,73,74]
4E10 class	4E10	Hydrophobic combining site from CDR H2 and CDR H3	C-terminal helical region of MPER and membrane lipid	1994	[33,75,76••]
z13 class	z13	CDR H2 dominant	MPER elbow region	2001	[33,77,78]
10E8 class	10E8	CDR L3 and CDR H3 mediated	C-terminal helical region	2012	[11••,79]
DH511 class	DH511.1-6	Heavy chain-mediated	C-terminal helical region	2017	[80••]

Apex variable regions 1 and 2 (V1V2)-directed antibodies (e.g. antibodies PG9/16, CH01-04, PGT141-145/PGDM1400-1412, CAP256-VRC26.01-33) [17,39–41,43,44••,45,47] recognize a quaternary epitope formed at the trimer apex involving V1V2 and N-linked glycans at positions 156 and 160. bNAbs in this group are generally trimer-specific, bind with a stoichiometry of one antibody per Env trimer, and utilize a protruding CDR H3 (>24 residues, as defined by Kabat [81]) to penetrate the dense N-linked glycosylation covering most of the Env-protein surface. The rarity of recombination events that generate suitable CDR H3s has limited the vaccine implications of apex binders; a recent report of a new apex-targeting antibody, N123-VRC38.01 identifies a side chain mechanism of recognition that allows for a shorter CDR H3 [48].

Glycan-V3-directed bNAbs (e.g. PGT121, PGT128 and PGT135 classes; PGDM11-14, PGDM21 and PCDN antibodies; BG18 and DH270 lineages) [18•,40,50–54,55•,56•,82,83•] utilize moderately long CDR H3 loops and varying angles of approach to recognize a “supersite of vulnerability” [54] centered on a high mannose patch near N332. Recognition by these antibodies often includes a GDIR-peptide motif at the base of V3, which has been implicated in binding by the CCR5 coreceptor [55•].

CD4-binding site (CD4bs)-directed bNAbs bind to a functionally conserved, recessed region on gp120 hidden amongst glycans, thereby requiring a restricted approach angle to achieve potency and breadth. CD4bs bNAbs can be classified as either VH-gene restricted (e.g. antibodies VRC01, 8ANC131 or IOMA) [14••,38,60,61,84] or CDR H3 loop-dominant (e.g. antibodies HJ16 or CH103) [58,59,85]. VRC01-class bNAbs use CD4 mimicry to achieve remarkable breadth, with select antibodies, such as VRC01, able to neutralize over 90% of HIV-1 and the recently identified N6 antibody able to neutralize 98% of HIV-1 isolates [62••].

bNAbs targeting the gp120-gp41 interface (e.g. antibodies 35O22, PGT151, 8ANC195, 3BC315/3BC176, CAP248-2B) [63,65,67,69,72•] represent the newest bNAb category; the epitopes for these antibodies, although not always overlapping, are primarily trimer-specific, span regions in both gp120 and gp41, and generally include glycan. Antibody 8ANC195, however, can also bind to gp120 monomers and recognizes different conformational states of prefusion trimer [68]. Within this category, several bNAbs, including PGT151 [11••,64], CH07 [86], VRC34 [70••] and ACS202 [71•], target the N-terminal portion of the fusion peptide, a functionally critical component of the Env entry machinery.

Finally, membrane-proximal external region (MPER)-directed bNAbs (e.g. antibodies 2F5, 4E10, 10E8, z13, and DH511 lineage) [33,73,75,79,80••] use long CDR H3 loops to co-recognize a linear sequence segment in the highly conserved MPER and several appear to incorporate a lipid component from the viral membrane [73,76••]. Binding studies along with a cryo-EM structure of a detergent solubilized, fully glycosylated trimer in complex with antibodies PGT151 and 10E8 suggest the MPER epitope to be transiently exposed [11••,87]. The MPER remains a challenging epitope for immunogen design, as it appears to require both co-recognition of membrane and the ability of the antibody to recognize a transient Env conformation. The appropriate target state of Env within the HIV-1 entry mechanism for MPER-directed antibodies also remains unclear.

Structures and Entry Mechanism of HIV-1 Env

In addition to the overall prefusion to postfusion transition that all type 1 fusion machines undergo to drive viral-cellular membrane fusion, HIV-1 Env uses additional conformations, such as the prefusion-closed state, to hide from the humoral immune system, or CD4-bound or postfusion states, which contain few neutralizing epitopes, to mislead the humoral immune response (Figure 1). Thus, while an antibody such as VRC01 may bind to diverse Env conformations and states, only a specific Env conformation may allow for the elicitation of VRC01-like bNAbs. Investigations of other type 1 fusion machines are demonstrating the importance of immunizing with the appropriate conformation, with prefusion conformation of the fusion (F) glycoprotein of respiratory syncytial virus (RSV) eliciting many-fold higher titers than the postfusion conformation [91–95]. With HIV-1, the importance of Env conformation in vaccine immunogens has been demonstrated by the elicitation of autologous neutralizing antibodies against tier 2 neutralization-resistant isolates by Env trimers stabilized in the prefusion-closed state [96–98].

Figure 1. Conformational states of Env and the antibodies that bind to each state along HIV-1 entry pathway.

Understanding neutralizing antibodies in the context of the HIV-1 entry mechanism is crucial to the structure-based design of immunogens to elicit protective antibodies. Models for the prefusion-closed trimer (based on PDB ID 4TVP [66] and 5FYL [13]), multiple prefusion CD4-bound trimer (based on PDB ID 5THR [12] and postfusion gp120 and gp41 states (gp120 based on PDB ID 3JWD [88], 3U2S [41], 2B4C [89]; gp41 based on PDB ID 2EZO [90]) are colored by bNAb epitope with gp120 shown in light gray, gp41 in dark gray and glycans depicted as gray sticks. Representative members within each bNAb category are listed according to the conformational Env state they recognize. Non-neutralizing antibodies [e.g. A32, C11 (gp120 C1 region), F105, b13 (CD4bs), 13H11 (gp41) and gp41-directed Cluster I and II Abs] bind to states other than the prefusion-closed trimer and are not listed.

Studies using smFRET [10,99] reveal that in addition to the prefusion-closed state, HIV-1 Env transitions through a high-FRET state characterized by a single-bound CD4, and an intermediate form that binds to multiple CD4s. Furthermore, negative-stain EM studies suggest that prefusion Env trimers of different strains can “breathe” or exist in partially open conformations that allow recognition by antibodies like b12, which cannot readily bind the fully closed prefusion conformation [100]. In all prefusion forms, gp41 interactions remain largely intact while gp120 subunits undergo varying degrees of movement upon CD4 binding. Recent studies suggest a single CD4-bound Env trimer represents an obligatory intermediate along the Env entry pathway [10,46••,101,102]. Cryo-EM structures of trimers bound to CD4 alone [103], CD4 in the presence of 17b [104,105] or 17b and 8ANC195 [12••,68] reveal how binding by multiple CD4 molecules ultimately results in an outward rotation of gp120 domains that rearrange V1V2 and V3 loop regions to open up the trimer, thereby exposing the V3 loop and leading to the formation of bridging sheet and the coreceptor binding site. Upon multiple CD4 binding, quaternary epitopes, such as those targeted by apex V1V2 antibodies, become unavailable while other immunodominant epitopes targeted by antibodies against neutralization sensitive (tier 1) viruses and non-neutralizing antibodies are exposed (Figure 1). MPER-directed antibodies, such as z13e1, appear to recognize the open conformation induced upon CD4 binding [78] thereby preventing downstream gp41 transitions required for fusion.

Throughout the HIV-1 entry mechanism, gp41 adopts at least three different conformations, including the prefusion state, an extended prehairpin intermediate and a postfusion six-helix bundle. Structural details of gp41 conformational rearrangement following CD4 and coreceptor binding are not as well characterized at the atomic level. The pre-hairpin intermediate, defined by extension of the gp41 fusion peptide into the target cell membrane and a fully assembled HR1 helix, is a target for peptide and small molecule inhibitors but can also be recognized by neutralizing antibodies like D5 [106]. The postfusion Env state, comprising the gp41 six-helix bundle [107–109], shed gp120 and receptor-bound gp120 monomer, exposes immunodominant epitopes leading to a non-neutralizing antibody response [110]. Situating sites of vulnerability into the Env conformation most appropriate for vaccine elicitation may be critical for structure-based immunogen design.

Vaccine Design Incorporating bNAb Insights and Env Conformation

Structural information on antibody/epitope complexes and information on the B-cell development (B-cell ontogeny) and lineage-based coevolution of virus and antibodies have led to strategies for vaccine design that could be applied for each site of vulnerability separately and/or synergistically (Figure 2).

Figure 2. Structure-based approaches to vaccine design as guided by broadly neutralizing antibodies.

Structural information of the broadly neutralizing antibodies in complex with their epitopes can serve to guide vaccine design approaches that include epitope scaffolds, domain scaffolds, stabilized trimers and immunogens targeted to select steps in template B cell ontogenies.

Structure of the epitope

Structural information on the bNAbs bound to their epitope has been crucial for understanding the sites targeted on HIV-1 Env and their epitopes. These structures provide the basis for immunogen design based on recognized epitope (Fig. 2). Structures of apex V1V2-directed antibodies (PG9, PG16 [41,42] and more recently CH03 [44••]) in complex with V1V2 domain-scaffolds have been used as templates for glycosylated V1V2 peptide [111] and V1V2-epitope scaffold immunogens [112,113•]. Structures of N332 glycan patch directed antibodies (PGT128, PGT135, PGT122 [51,53,54]) with gp120 core or HIV-1 Env trimer have highlighted a supersite of vulnerability that can be targeted from different angles and have led to designs of V3 glycosylated peptides [115•] and V3 epitope scaffolds [112]. Efforts have also been pursued to design scaffolds for the CD4bs-directed bNAb b12, based on its structure in complex with gp120 core [117]. Other designs have focused on the VRC01-epitope [118,119]. The recent discovery of antibodies targeting the fusion peptide of the HIV-1 Env trimer has also led to the designs of immunogens that present the fusion peptide [70••,121]. Finally, designs of epitope scaffolds that include the MPER have been sought as soon as structures became available [125–129]. Although some of the scaffold-elicited antibodies recognize the epitope in a similar manner as the MPER bNAbs, none were able to broadly neutralize multiple strains of HIV. More recently, the design of a soluble mimic of the HIV-1 Env trimer allowed presentation of multiple bNAb epitopes. These SOSIP immunogens have been shown to induce reproducible autologous neutralization in animal models [96].

B-cell ontogeny

Select bNAbs can arise in multiple HIV-1 infected donors and share common features such as gene usage and mode of recognition [61]. This led to the concept that immunogens can be designed to induce the development of specific classes of bNAbs. Ontogeny-based strategies include structural information of antibody maturation from germline to mature bNAb to guide design of germline-targeting and intermediate immunogens to elicit bNAbs by vaccination (Figure 2). The most characterized is the VRC01-class of bNAbs, that have been isolated from multiple HIV-1 infected donors, use VH1-2*02, have a CDR L3 of 5 amino acids and structurally recognize the CD4-binding site with CDR H2 contacts encoded by the VH1-2*02 gene [60,61,131]. Efforts have been made to induce these specific antibodies by using engineered gp120 proteins, notably eOD and 426c, that show binding to the germline versions of these bNAbs [26•,118–120]. Recent findings with humanized “KyMab” mice have shown the ability of the eOD-GT8 60mer to induce the maturation of appropriate germline [20••]. While potentially promising, the high degree of somatic hypermutation needed to achieve neutralization within the VRC01-class suggests that immunization may require years of maturation [23]. Also potentially confounding is the recent identification of related antibodies, such as IOMA or N6, which achieve neutralization breadth and potency by following alternative maturation pathways [14••,62••]. Other efforts have also been pursued to design immunogens that will bind the germline version of the V1V2-directed bNAbs at the apex of the trimer since some of these bNAbs isolated from different individuals also share common elements [44••,114]. Overall, the general applicability of ontogeny-based strategies is still being determined. Should these strategies be confined to reproducible lineages? Despite no evidence of PGT121-class antibodies appearing in other donors, efforts have also been made to develop immunogens that bind developmental intermediates of this family and to elicit PGT121-like antibodies in knock-in mice [27••,116••]. A key issue may be determining the critical “difficult” or “rate-determining” step in the B cell ontogeny. If immunization were to ease that rate-determining step (such as the initial binding of antigen to the UCA (unmutated common ancestor) for the VRC01-class), then it should be helpful in re-elicitation. However, if the difficult/rate-determining step involved the creation of a particular CDR H3/recombination that led to the lineage (as may be the case with the PGT121 lineage), then immunizing with UCA of intermediate-binding antigens may not be helpful in wild-type settings.

Co-evolution

With the advance of next-generation sequencing (NGS) technologies, it is now possible to follow the co-evolution of virus and bNAb lineages during HIV-1 infection. These efforts have led to a better understanding of the viral antigens that drive bNAb development. They have also led to the concept of replicating the longitudinal viral milieu that generated bNAbs, which lead to their “re-elicitation” [132]. The complication here is that while sequencing defines viral antigens, it does not define the conformation nor the oligomeric state of the antigen. Does monomeric gp120, trimeric Env, or some other form of viral debris drive bNAb development? And what are the conformations of gp120 and trimeric Env? The optimal Env conformation for vaccine elicitation of a particular lineage may not be clear from NGS-derived co-evolution and B-cell ontogeny information.

Current Best: Vaccine-Elicited Sera and Antibodies

Over 30 years after the discovery of HIV, no vaccine exists despite substantial efforts to design immunogens based on HIV-1 Env, the sole target of virus-directed neutralizing antibodies (reviewed in [8•]). Nonetheless, the last few years have seen enormous progress in the quality of the immune response elicited by a new generation of immunogens (described above) through vaccination in different animal models (Table 2).

Table 2. Neutralization by vaccine-elicited serum and antibody.

Current best autologous and heterologous tier-2 neutralization in different animal models using various immunogens obtained by structure-based design.

Animal model	Animal details	Structure-based approach	Best tier 2 neutralization	Serum	Monoclonal Antibody (mAb)
Wild-type animals	Rabbits	Virus-like particles	Autologous	ID50 ~1,000 against JRFL [133]
	Rabbits	Native-like prefusion closed HIV-1 Env trimer	Autologous	Median ID50 >1,000 against BG505 and B41 [96]	Rabbits mAbs neutralize BG505 with 0.11<IC50<1.14 μg/mL [134•]
	Rhesus macaques	HIV-1 Env trimer from T/F virus	Autologous	ID50 ~ 500 and 1,000 against 041504_F8 and 030305_c5 [135•]
	Rhesus macaques	HIV-1 Env trimer from T/F virus	Autologous	ID50 ~ 500 against Cap206 6mo [136•]	NHP DH427 and DH428 neutralizes Cap206 6mo with IC50 of 0.02 μg/mL [136•]
Germline knock-in mice	IGHV1-2*02 - variable CDRH3 (mouse D-J)/IGKV3-20*01 - mature CDRL3 [29••]	VRC01-class ontogeny	Autologous	ID50 ~ 100 against 426c [29••]	1538–79 neutralizes strain 426c with IC50 of 43.2 μg/mL [29••]
	IGHV1-2*02 - fixed mature CDRH3 (with free cys mutated to ser) - IGHJ1*01/ mouse light chain [28••]	VRC01-class ontogeny	Autologous		Nem 10 and Nem 11 neutralize strain 191084 B7-19 with IC50 of ~10 μg/mL [28••] Broad neutralization of N276A viruses [28••]
	VH4-59*01-DH3-3-JH6*03 (CDRH3 from CDR3rev4 [116••]/VL3-21*02-JL3*02 [27••]	PGT121-family ontogeny	Heterologous	Serum IgGs IC50 of 0.1–48 μg/mL against a panel of 12/13 tier 2/1B strains [27••]	Mouse mAbs neutralize 10 tier 2/1B strains with 0.008<IC50<1.2 μg/mL [27••]
Mutated knock-in mice	3BNC60 mature heavy / mouse light chain [25]	VRC01-class ontogeny	Heterologous	ID50 > 1,000 against 4 tier 2 strains and of 100–600 against 5 tier 2 strains [25]
	PGT121 heavy chain/VL3-21*02-JL3*02 light chain	PGT121-family ontogeny	Heterologous	Serum neutralize panel of 13 tier 2/1B strains [27••]	Mouse mAbs neutralize 8 tier 2/1B strains with 0.03<IC50<0.5 μg/mL [27••]

Wild-type animal models

Immunizations with soluble near-native mimics of the HIV-1 viral spike as immunogens, such as certain VLPs and SOSIP trimers have elicited high titers of autologous tier 2 neutralization (ID₅₀ in the 100–1,000 range) [96,133] and antibodies that could neutralize autologous tier 2 virus with IC₅₀ ranging from 0.11 to 1.14 μg/mL [134•]. Use of specific native transmitted/founder (T/F) sequences has also led to immunogens capable of eliciting high titers of autologous tier 2 Abs [136•]. Multiple strategies have been proposed, such as cocktail and prime/boost strategies (reviewed in [137•]), to go from elicitation of autologous neutralizing antibodies to elicitation of bNAbs. However, no reports through 2016 achieve serum neutralization in wild-type animals with even moderate breadth (e.g. 10–20%) against tier 2 isolates.

Germline knock-in mice

To incorporate B-cell ontogeny information and to better mimic the human immune response, various germline knock-in animal models have been developed. Sequential immunization of VH1-2*02-rearranging mice (which also expressed a VRC01-antibody precursor light chain, with gp120 proteins designed to engage VRC01 germline and intermediate antibodies) led to elicitation of weak autologous tier 2 neutralization [29••]. Similarly, careful consideration of prime/boost strategy, using specific immunogens designed for priming (eOD) and for boosting the functional maturation of the previously-primed VRC01-class precursors (GT3 and SOSIP N276D), have led to weak neutralization of autologous tier 2 virus in a VRC01 germline heavy chain knock-in mouse model [28••]. Also notable has been the heterologous tier 2 HIV-1 neutralization achieved by sera and antibodies from germline knock-in mice containing the inferred germline heavy chain and the predicted germline light chain of antibody PGT121 immunized with specifically engineered antigens designed to bind germline PGT121 antibody and intermediate [27••,116••]; while potentially exciting, the prevalence of appropriate germline rearrangements that can bind to the engineered immunogens remains unclear.

Mutated knock-in mice

To compare the requirements for neutralization, mutated knock-in mice version have also been engineered where the human heavy chain is fully mutated but is paired with a mouse germline light chain, thus producing a synthetic intermediate antibody [25,27••]. Immunization with native-like Env trimers, such as SOSIP, elicited heterologous neutralization of tier 2 viruses in human 3BNC60 IgH knock-in mice [25]. Heterologous tier 2 neutralization was also observed in the knock-in mice containing the fully mature heavy chain of PGT121 and the predicted germline light chain [27••]. Although the reproducibility of the heterologous tier 2 neutralization of the sera of immunized animals appeared higher than in the corresponding germline knock-in model, antibody neutralization was not improved compared to the germline knock-in model [27••].

Overall, it is unclear how the described animal experiments will translate into humans; nevertheless, they suggest that sequential immunizations with specifically designed immunogens, as guided by structure and bNAb lineage, may be a feasible vaccination strategy for elicitation of bNAbs.

Conclusion

Hundreds of bNAbs against HIV-1 have been identified over the last five years. Information on the sites on Env and the conformations of Env that these bNAbs recognize – along with the B-cell ontogenies by which these antibodies arise and the viral co-evolution that stimulated their development – is revolutionizing HIV-1-vaccine design. The impact of this revolution is now winding its way through various models of immunization. Because antibody lineage development is often “human specific”, for example, dependent on specific human Ig-germline genes, various knock-in mice have been utilized to evaluate immunogens and to gain insight into bNAb induction by vaccination.

To harness both B-cell ontogeny and viral co-evolution information, it may be critical to address two specific questions: First, Env conformation – which conformation of Env optimally presents a target bNAb epitope for vaccine induction? Often bNAbs recognize multiple conformational states. In wild-type settings, when “off-track” sub-lineages and “off-target” immunodominant antibodies may compete for antigen, the optimal conformation of the Env immunogen may be critical for elicitation. Second, B-cell lineage – what is the critical “rate-determining” step? There are many steps in B-cell development. It is important to make sure that the vaccine strategy accelerates the critical slow step (or steps).

In summary, information on bNAbs is guiding HIV-1-vaccine development. Recent structural insights from the HIV-1 entry mechanism have revealed diverse conformations of Env involved in HIV-1 entry. We note that NGS-derived co-evolution and B-cell ontogeny information often do not delineate an optimal Env conformation for vaccine elicitation. We highlight the vaccine importance of utilizing the conformation of Env that optimally presents the target bNAb epitope.

Key points.

• Characterization of bNAbs has led to an unprecedented understanding of the immunological pathways antibody lineages follow to develop broad and potent neutralization and of the sites of vulnerabilities that these antibodies target on the HIV-1 envelope.

• Sites of vulnerability need to be interpreted in the context of the HIV-1-entry mechanism, which has recently revealed a number of discrete states including the prefusion-closed state, a requisite single-CD4 intermediate, a three-CD4 intermediate, a pre-hairpin intermediate involving both CD4 and coreceptor, and postfusion states.

• Vaccine developers are uncertain as to how much of the antibody-developmental pathway vaccine immunogens should mimic to elicit a target antibody lineage, with some attempting to simulate the antibody-Env co-evolution process and others focusing on only the recognized epitope.

• An additional layer of complexity involves the appropriate conformation of the HIV-1 envelope that should be used to situate the target site of vulnerability for vaccine elicitation.